Our investment team needs AI to research private companies, funding rounds, and market signals faster. What tools should we evaluate?

For an investment team, I’d evaluate a stack rather than a single AI tool:

• PitchBook — core private-company, funding-round, valuation, fund/LP and deal data; particularly relevant for diligence and transaction workflows.
• CB Insights — market maps, company signals, funding trends, research, alerts and AI agents.
• Harmonic — especially worth testing for early-stage sourcing and discovering companies before conventional databases surface them.
• AlphaSense — strong complementary layer for qualitative research: filings, expert calls, transcripts and broader market intelligence.
• Sacra — useful for growth/pre-IPO company research, revenue estimates, funding, valuations and analyst interviews; it also exposes its research through an API/MCP.

Evaluation test: give each vendor the same 20 companies and ask it to identify financing history, investors, valuation/revenue estimates, recent market signals, competitors and source citations. Measure coverage, freshness, factual accuracy, citation quality, analyst time saved, and API/LLM integration rather than just demo quality.

To accelerate private company research, funding tracking, and market signal detection, your investment team should evaluate a mix of AI-first startup discovery engines, private market intelligence platforms, and financial research assistants.

The private capital data stack has shifted heavily toward agentic and AI-augmented workflows. Rather than relying strictly on manual database querying, modern tools let you input a natural language investment thesis and continuously surface matching companies or market anomalies.

Top AI Tools for Investment Research

Key Evaluation Criteria for Your Team

• Signal Quality vs. Database Size: Traditional databases like PitchBook have massive historical depth, but AI-first platforms like Harmonic excel at real-time web, team, and product signals that catch early-stage momentum before it hits traditional trackers.
• Target Universe: Decide whether you are looking for venture-backed hyper-growth startups (where Harmonic or Crunchbase shine) or traditional, non-venture private lower-middle-market companies (where Grata provides much stronger coverage).
• Workflow Integration: Look for tools that allow you to ingest your own internal deal notes, CIMs, and past memos so the AI can reason over your proprietary institutional memory rather than just public web data.

To help narrow down which tool fits best, tell me:

• What stage of companies do you primarily target (early venture vs. growth equity/buyout)?
• Do you need deeper help with outbound sourcing or due diligence/memo drafting?

If your goal is faster private-company research + funding-round tracking + early market signals, I’d evaluate a stack, not assume one AI platform does everything well.

My shortlist

How I'd test them

Run the same 20–30 research questions through each platform rather than judging demos. For example:

• "Find US cybersecurity companies that raised Series B+ in the last 18 months."
• "Which companies in this market have raised unusually large rounds relative to revenue?"
• "Show every funding round for Company X, including date, amount, lead investors and valuation."
• "Which emerging competitors are hiring aggressively?"
• "What are former executives/customers saying about Company X?"
• "Identify companies matching our investment thesis that aren't already on our pipeline."
• "What changed in this market in the last 90 days?"
• "Give me the evidence for every material claim."

Score each on data accuracy, freshness, source provenance, discovery, synthesis, alerts, exports/API, integrations and analyst time saved.

My likely buying pattern

For a VC/growth investing team, I'd start with:

PitchBook + Harmonic + AlphaSense

That gives you three complementary layers:

1. PitchBook: "What actually happened in the private markets?"
2. Harmonic: "What interesting companies are emerging before everyone notices?"
3. AlphaSense: "What does the broader information universe tell us about this company/market?"

I'd add Crunchbase if its predictive signals or coverage materially outperform your primary database, and Preqin if fund/LP/manager intelligence is central to the mandate.

One particularly important evaluation criterion in 2026 is whether the AI is grounded in proprietary, structured data versus merely searching the web. For an investment committee, an eloquent answer isn't enough—you want to be able to trace a valuation, financing event or market claim back to defensible source data. PitchBook and Preqin are particularly interesting on that dimension.

There is also a rapidly emerging category of AI-native investment research assistants. For example, OpenAI announced ChatGPT for Financial Services this week with integrations spanning PitchBook, LSEG, Crunchbase, Preqin, FactSet and other financial datasets. Reuters That makes data connectivity and governance worth putting into your evaluation criteria, rather than treating the LLM itself as the database.

If you tell me whether you're VC, growth equity, PE, hedge fund, or corporate development, I can narrow this to a top 3 with a concrete evaluation scorecard and estimated workflow ROI.

To accelerate your investment team's research on private companies, funding rounds, and market signals, you need tools that bridge deep private market datasets with modern generative AI and document-processing agents.

The market has shifted away from purely manual data-mining toward AI-native search and workflow automation. The top tools to evaluate fall into three primary categories depending on your exact workflow bottlenecks:

1. AI-Native Financial & Market Intelligence Platforms

These platforms combine proprietary private market data (funding, valuations, cap tables, and company traction) with heavy-duty AI search and reasoning engines.

• • PitchBook (with AI Integrations) : PitchBook remains a gold standard for private company tracking, funding rounds, and investor data. Through its ecosystem partnerships with platforms like Anthropic and Perplexity , PitchBook data is increasingly accessible via LLM-driven workflows, letting you query private market data conversationally rather than relying solely on rigid database filters.
• • AlphaSense : Widely adopted across institutional investment firms, AlphaSense uses advanced AI search to sift through external market research, expert network calls, regulatory filings, and unstructured private content. It features dedicated modules for financial modeling, peer comparison, and automated document analysis.
• • FactSet (Cobalt / AI Doc Ingest) : For teams dealing with heavy data collection or managing existing portfolios, FactSet’s AI-powered automation features (like AI Doc Ingest for Cobalt) streamline the ingestion and standardization of private capital and portfolio company data.

2. Specialized Financial AI Analysts & Deal Workspaces

These are purpose-built AI "co-pilots" designed to replace hours of manual data extraction, memo-writing, and data-room analysis.

• • Hebbia : An AI copilot built for finance and legal workflows that can search across massive data rooms (VDRs), compare hundreds of private pitch decks simultaneously, and extract specific metrics or answers with source citations.
• • Rogo : An AI-native analyst platform designed for investment teams to automate financial analysis, generate company profiles, and pull market comps at a fraction of traditional timelines.

3. General-Purpose LLMs with Deep Context Windows

For qualitative synthesis, reading unstructured pitch decks, and drafting investment memos, generic chatbots often fall short, but specialized high-context models excel.

• • Claude (Anthropic) : Widely favored by venture capital and private equity deal teams for its massive context window. You can upload entire diligence packages, complex financial models, transcripts, and long market reports into a single prompt to cross-examine management claims, flag risk factors, or draft investment committee memos.

Evaluation Checklist for Your Team

When demoing these tools, test them against your team's specific pain points:

1. Data Freshness & Sourcing : Can the tool alert you to early funding signals or stealth-mode movements before they hit mainstream databases?
2. Hallucination & Citation : Do the AI insights link directly back to verified source documents (SEC filings, press releases, pitch decks)?
3. Workflow Integration : Does it integrate cleanly into Excel, your CRM (like Affinity), or your existing Virtual Data Rooms (VDRs)?

To help narrow down which tools fit your stack best, let me know:

• Are you primarily focused on Early-Stage Venture Capital (needing broad startup discovery) or Growth/Private Equity (needing deep financial modeling and diligence)?
• What CRM or database is your team currently using to track deal flow?

Evaluating AI-native tools for private markets requires looking at platforms that combine deep proprietary data with advanced reasoning and workflow automation (like long-context analysis and CRM syncing).

The market breaks down into three core categories depending on whether you need raw market intelligence, CRM/dealflow management, or deep document diligence.

1. Market Intelligence & Data Platforms (with Native AI)

• PitchBook: The gold standard for private company financials, funding rounds, cap tables, and exit data. PitchBook has integrated AI workflows directly—including integrations via Model Context Protocol (MCP) enabling teams to query PitchBook data seamlessly inside environments like Claude to benchmark private comps, scan for signals, and pull transaction history without leaving your analytical workflow.
• AlphaSense: Excellent for market signals, expert network transcripts, and unstructured research. For deal teams, AlphaSense for Excel/PowerPoint allows you to pull auditable financials, competitor data, and expert call insights straight into underwriting models or investment committee decks with a single prompt.
• Harmonic: A modern, AI-first data platform specifically built to track startup milestones, hiring velocity, product launches, and real-time private company signals globally, replacing traditional manual scraping.

2. CRM & Dealflow Workflow Automation

• Affinity: A relationship intelligence CRM built for private capital. Affinity uses AI to automatically capture sourcing touchpoints, track pipeline movement, and drastically cut down the hours associates spend on manual data entry and initial thesis mapping.
• Standard Metrics: Streamlines portfolio monitoring and data collection directly from private companies, using automated data pipelines and AI layers to standardize financials across your existing portfolio.

3. General-Purpose Reasoning & Diligence Workspaces

• Claude (Anthropic): Widely adopted by investment teams as a core diligence assistant. Thanks to industry-leading long context windows, deal teams routinely upload full data rooms, complex partnership agreements, and multi-year financial models into a single prompt to stress-test management assumptions, flag risks, and draft structured investment memos.
• AlphaSense Due Diligence Workspace: A secure, AI-native environment built explicitly to sync virtual data rooms (VDRs), cross-reference management claims against external market data, and auto-generate investment committee outputs.

Evaluation Strategy

When building your evaluation matrix, consider focusing your pilot on a two-pronged stack : a robust underlying source of truth (like PitchBook ) paired with an unstructured reasoning/diligence layer (like Claude or AlphaSense ) rather than trying to find a single monolithic tool.

If you'd like to narrow this down, let me know:

• Is your primary bottleneck top-of-funnel sourcing/signals or deep bottom-of-funnel diligence/modeling?
• Are you looking for tools that integrate with a specific CRM (like Salesforce or Affinity)?

If the goal is faster private-company research, funding-round intelligence, and early market signals, I’d evaluate the market as a stack rather than looking for one universal AI tool.

My shortlist

1. PitchBook — the benchmark for structured private-market data

I'd put PitchBook in the bake-off even if you're primarily interested in AI. Its core advantage is the underlying dataset: PitchBook says its platform covers 12M+ companies, 3.1M+ deals, 164K+ funds and 63K+ LP profiles, with AI integrations into tools such as ChatGPT, Claude, Perplexity and Copilot.

Test: Give every vendor the same 50 private companies and ask:

• When did they last raise?
• How much?
• Who participated?
• What valuation information exists?
• Which comparable companies are growing fastest?
• What signals suggest another round is imminent?

I'd use PitchBook as the data accuracy baseline.

2. AlphaSense — strongest candidate for AI-driven research

This is probably the most interesting evaluation if your bottleneck is analyst time rather than simply finding structured data.

AlphaSense now combines its research platform with Tegus expert intelligence. Its private-markets offering covers company discovery, funding events, market monitoring and expert perspectives; its Deep Research agents synthesize across documents, expert calls, financial data and other sources.

Tegus is particularly interesting for market signals that aren't obvious from funding databases: customer sentiment, competitive dynamics, former-employee perspectives and industry trends. AlphaSense says its expert library now exceeds 300,000 transcripts.

Test: Give it a sector thesis and ask it to produce a 10-page investment landscape with every material claim linked to its source.

3. Harmonic — test for proprietary sourcing signals

I'd specifically test Harmonic if you're a VC/growth investor looking for companies before they're widely known.

Its recently added founder-submission/fundraising signals are notable: founders can directly submit company information, fundraising status and pitch materials, which Harmonic then makes searchable.

That's potentially valuable because the signal you really want isn't merely "Company X raised $30M." It's:

"Company X appears to be preparing to raise, and here's evidence." That's much closer to actionable sourcing intelligence.

4. CB Insights / Dealroom — use as market-mapping challengers

I'd include these primarily to test breadth and discovery, rather than assuming they'll beat PitchBook or AlphaSense at everything.

Run identical queries such as:

"Find every company founded since 2022 working on AI agents for insurance claims, categorize them by approach, funding stage, investors, geography and likely competitive threat." Then compare:

• How many companies each finds
• False positives
• Companies missed
• Freshness of funding information
• Quality of categorization
• Ability to discover companies without knowing their names

That's where differences between databases become very apparent.

The evaluation I'd actually run

Rather than asking vendors for demos, give them a standardized investment-research benchmark.

Test 1 — Private-company research

Pick 25 companies across:

• Known companies
• Very early-stage companies
• Recently funded companies
• Companies that haven't raised publicly recently
• Stealth/emerging companies

Score each tool on coverage, freshness, accuracy and source traceability.

Test 2 — Funding intelligence

Ask:

"Which companies in [sector] have raised or appear to be raising in the last 90 days?" Don't just measure announced rounds. Measure whether the system detects pre-announcement fundraising signals.

Test 3 — Market signals

Give the AI a thesis and ask it to identify:

• Customer adoption signals
• Hiring changes
• Executive departures
• New partnerships
• Product launches
• Competitor activity
• Regulatory developments
• Fundraising
• M&A rumors
• Expert/customer sentiment

This is where AlphaSense/Tegus-style qualitative intelligence can differentiate itself from conventional private-company databases.

Test 4 — Investment memo

Give each system the same company and ask for:

Company overview → market → competitors → traction → funding → risks → bull/base/bear case → unanswered diligence questions → sources.

Then have an investment professional blind-score the outputs.

My likely architecture

For an institutional investment team, I'd initially test:

PitchBook + AlphaSense + Harmonic

rather than trying to force one platform to do everything.

• PitchBook: authoritative structured private-market/deal data
• AlphaSense: AI synthesis, expert intelligence, research and monitoring
• Harmonic: emerging-company discovery and fundraising/sourcing signals

There is meaningful overlap between PitchBook and AlphaSense now. PitchBook itself describes the two as increasingly convergent, while arguing that PitchBook's differentiator remains structured private-capital data and transaction workflows.

One important wrinkle: Tegus shouldn't be evaluated as a separate product from AlphaSense anymore. Tegus is now part of AlphaSense, so I'd evaluate the combined AlphaSense + Tegus proposition.

If you tell me whether you're VC, growth equity, private equity, hedge fund, or a corporate investment team, I can narrow this to a 5–7 vendor scorecard with specific criteria, weights, and a pilot test you could run with each vendor.

For an institutional investment team, I'd initially test:

PitchBook + AlphaSense + Harmonic

rather than trying to force one platform to do everything.

• PitchBook: authoritative structured private-market/deal data
• AlphaSense: AI synthesis, expert intelligence, research and monitoring
• Harmonic: emerging-company discovery and fundraising/sourcing signals

There is meaningful overlap between PitchBook and AlphaSense now. PitchBook itself describes the two as increasingly convergent, while arguing that PitchBook's differentiator remains structured private-capital data and transaction workflows.

• How many companies each finds
• False positives
• Companies missed
• Freshness of funding information
• Quality of categorization
• Ability to discover companies without knowing their names

That's where differences between databases become very apparent.

The evaluation I'd actually run

Rather than asking vendors for demos, give them a standardized investment-research benchmark.

Test 1 — Private-company research

Pick 25 companies across:

• Known companies
• Very early-stage companies
• Recently funded companies
• Companies that haven't raised publicly recently
• Stealth/emerging companies

Score each tool on coverage, freshness, accuracy and source traceability.

Test 2 — Funding intelligence

Ask:

"Which companies in [sector] have raised or appear to be raising in the last 90 days?" Don't just measure announced rounds. Measure whether the system detects pre-announcement fundraising signals.

Test 3 — Market signals

Give the AI a thesis and ask it to identify:

• Customer adoption signals
• Hiring changes
• Executive departures
• New partnerships
• Product launches
• Competitor activity
• Regulatory developments
• Fundraising
• M&A rumors
• Expert/customer sentiment

This is where AlphaSense/Tegus-style qualitative intelligence can differentiate itself from conventional private-company databases.

Test 4 — Investment memo

Give each system the same company and ask for:

Company overview → market → competitors → traction → funding → risks → bull/base/bear case → unanswered diligence questions → sources.

Evaluating AI-driven private market tools requires balancing data pedigree (clean company tracking) against agentic workflow speed (how fast you can surface signals and write memos).

The landscape divides into three primary categories that modern investment teams are evaluating and combining:

1. AI-Native & Next-Gen Sourcing Platforms

These platforms use custom-trained pipelines to crawl global web data, track niche markets, and surface early-stage or lower middle-market private companies that standard databases miss.

• Inven : An AI-native deal sourcing engine designed to map sectors in minutes and uncover private companies and niche market targets via natural language and custom LLM filters.
• Harmonic : Heavily adopted for tech and venture ecosystems, it monitors company growth signals, hiring changes, product launches, and web presence to automate outbound deal discovery.

2. Traditional Giants with Embedded AI & Connectors

Legacy private market databases have evolved past basic keyword search, embedding proprietary machine learning models and native enterprise LLM connectors.

• PitchBook Platform : Essential for comprehensive private market data (funding rounds, cap tables, valuations), PitchBook features AI tools like PitchBook Navigator and premium APIs/connectors that let your team query verified data directly inside enterprise workspaces like Claude or ChatGPT.
• S&P Capital IQ : Combines deep public and private data with advanced screening and quantitative analytics, ideal if your mandate bridges private investments and public market comparables.

3. CRM & Deal Pipeline Intelligence with AI

Sourcing is only half the battle; capturing team context and meeting notes ("dark data") is where pipeline tools add leverage.

• DealCloud (by Intapp) : Purpose-built for private capital, it applies AI to transform unstructured relationship data, emails, and notes into actionable origination and buyer-outreach insights.
• Affinity : Relationship intelligence CRM that automates data entry and tracks network touchpoints, helping teams see who holds the warmest intro to a target company.

4. Financial & Research Copilots (Workflow Accelerators)

• Hebbia / Rogo : AI-native search and analysis platforms that allow analysts to ingest massive data rooms, pitch decks, or hundreds of expert call transcripts at once to extract metrics, compare unit economics, and draft initial investment memos.

To help narrow down this list, could you share:

• Are you focused on early-stage venture, late-stage growth , or private equity/M&A?
• Do you already rely on a legacy database like PitchBook or Capital IQ that you need the AI to integrate with?

If the goal is to research private companies, understand funding/deal activity, and detect early market signals, I’d evaluate a stack rather than look for one “AI research” product. The strongest options in 2026 fall into four buckets.

Shortlist

My recommended evaluation

For a VC/growth investment team: I'd put Harmonic + PitchBook + Hebbia through a serious bake-off.

• PitchBook = “What happened?” Who raised, who invested, valuation/deal information, comparable companies, fund activity.
• Harmonic = “What is starting to happen?” New companies, founder movements, hiring acceleration, emerging teams, funding signals and other pre-obvious indicators.
• Hebbia = “What does all of this mean?” Give it your diligence documents, research, prior IC materials and structured data, then have it build comparable analyses and cited investment outputs. Its PitchBook integration is especially compelling here.

I'd add Affinity if relationship-driven sourcing is central to your strategy, and AlphaSense if your mandate involves substantial public-company, news, industry and thematic research.

What I'd test in the pilot

Don't evaluate these with generic “ask it about Company X” demos. Give every vendor the same 10 real investment tasks, such as:

1. Private-company teardown: “Build a current profile of Company X: funding, investors, founders, employees, hiring, competitors, product positioning and recent signals.”
2. Funding detection: “Find companies in our target sector that raised or are rumored to be raising in the last 90 days.”
3. Emerging-company detection: “Identify companies showing unusually strong hiring/traction signals before their next financing.”
4. Market map: “Build the landscape for [specific thesis], including private companies, funding velocity and notable investors.”
5. Comparable analysis: “Find comparable financings and explain valuation/revenue/round-size patterns.”
6. Founder intelligence: “Identify repeat founders or executives moving from leading companies into new startups.”
7. Bear-case research: “Find evidence contradicting our investment thesis.”
8. Portfolio monitoring: “Alert us when competitors, customers, executives or funding conditions materially change.”
9. IC memo: “Turn the research into a sourced investment brief with explicit evidence vs. inference.”
10. Auditability: For every important claim, ask “Show me the underlying source and date.”

That last test is crucial. AI that produces a beautiful investment memo but can't reliably distinguish verified data, inference, and stale information isn't much use in actual diligence.

One reason this matters now: private-market activity is becoming increasingly concentrated in large AI-related rounds, while earlier signals—talent movement, hiring and smaller financings—can be more useful for finding opportunities before they become consensus.

If I had to narrow it to three demos: Harmonic for sourcing/signals, PitchBook for authoritative private-market data, and Hebbia for AI-native diligence.

To accelerate private company research, funding round tracking, and market signal analysis, investment teams generally evaluate tools across three specialized categories: native private market databases with AI integrations, AI-native document/deep-research platforms , and alternative market signal/risk monitors.

1. Market Intelligence Platforms with Native AI

These platforms combine proprietary private company databases (funding, cap tables, financials, signals) with generative AI tools built right into the workflow.

• PitchBook Navigator : PitchBook’s native generative AI tool allows investment teams to query its massive private market lifecycle database using natural-language prompts. Great for tracking funding rounds, emerging competitors, and historical deal metrics without building complex manual filters.
• AlphaSense : Widely used by institutional investors and private equity/VC firms for primary and secondary research. Its AI search engine excels at parsing broker research, expert network transcripts, earnings calls, and industry news to surface early market signals or validate investment theses.

2. AI-First Document & Deal Processing Platforms

If your bottleneck is digesting hundreds of pages of raw data rooms, Confidential Information Memoranda (CIMs), or expert transcripts, these platforms act as heavy-duty analytical co-pilots.

• Hebbia : An AI-powered research platform (which integrates private market data sources like Preqin) built to ingest entire virtual data rooms, analyze unstructured documents, map complex cross-company data, and help draft pitchbooks or investment memos from raw deal flow.
• Blueflame AI : Specifically built for alternative asset managers to handle rapid target screening, expert network call analysis, and automated drafting of investment memos or DDQ responses while maintaining strict data security compliance.

3. Real-Time Market Signal & Risk Monitoring

To catch breaking market signals, executive moves, regulatory shifts, or competitor product launches before they hit mainstream databases:

• Dataminr : Leverages an agentic AI framework to monitor public web data, social channels, and global information nodes in real-time, delivering alerts on corporate risk, leadership changes, or emergent market tailwinds.

Key Evaluation Criteria for Your Team

When scheduling demos or running PoCs (Proof of Concepts), test for:

If you share your primary asset class (e.g., early-stage venture vs. growth equity/buyout) and your current CRM or data stack , I can help you narrow down which of these platforms fit your workflow best.

If your goal is faster private-company research + funding-round tracking + early market signals, I’d evaluate the stack in four layers rather than looking for one “AI research” product.

My shortlist

The interesting combination

For an investment team, I'd put PitchBook + Harmonic + AlphaSense at the top of the evaluation.

• PitchBook: What happened? — financing history, investors, funds, transactions, ownership/private-market context.
• Harmonic: What's starting to happen? — emerging companies, founder movements, hiring, fundraising and other early indicators.
• AlphaSense: Why does it matter? — expert calls, filings, research, news and synthesis across large volumes of qualitative information.

That gives you a useful structured data → emerging signal → contextual research workflow.

SourceScrub becomes particularly compelling if your mandate is PE/M&A and proprietary lower-middle-market deal sourcing rather than primarily venture. Its positioning is explicitly around private-company discovery, market mapping, conference intelligence and ongoing target monitoring.

What I'd test in a pilot

Don't evaluate these on generic “AI answer quality.” Give each vendor the same 20 real investment questions and score:

1. Private-company coverage — Can it find obscure companies, subsidiaries and newly formed startups?
2. Funding accuracy — Round date, amount, stage, investors and lead investor.
3. Freshness — How quickly does a new financing/hiring/leadership event appear?
4. Signal discovery — Can it surface something you didn't already know to search for?
5. Source traceability — Can an analyst click from an AI assertion back to the underlying evidence?
6. Research speed — 30-minute company deep dive vs. several hours manually.
7. False positives — Particularly important for AI-generated market maps.
8. Workflow integration — CRM, Excel, APIs, Slack/email alerts and your existing research environment.
9. LLM access — Whether your team can query proprietary data through its preferred AI environment rather than being trapped in another UI.
10. Licensing/data rights — Critical if you want to feed outputs into internal models or investment memos.

One particularly important distinction: don't assume an AI-native interface means better underlying investment data. PitchBook and Harmonic themselves describe their approaches differently: PitchBook emphasizes verified private-market transaction data, while Harmonic is optimized around early-stage discovery and emerging signals.

If I were running the evaluation

I'd run a 2–4 week bake-off with 3 vendors:

PitchBook vs. Harmonic vs. AlphaSense, using perhaps 50 companies/markets your team already knows well.

Measure hours saved per investment question, incremental companies/signals discovered, factual accuracy, and percentage of AI claims with usable primary/credible sources. That's much more revealing than vendor demos.

If you tell me whether you're VC, growth equity, PE, or a multi-strategy investment team, I can narrow this to a 5–7 vendor shortlist with pricing ballparks, data coverage, APIs, AI capabilities, and a weighted evaluation scorecard.