VC Analyst

AgentLed research

AI agents in investment teams: buy, build, or combine?

A short exploratory study of how venture teams introduce AI into sourcing, diligence, portfolio work and firm operations.

How do teams decide which tools to buy, which capabilities to build, and where human judgment must stay in the loop?

Share your perspective

Participate to receive the findings brief and practical industry insights from peer investment teams.

Why this study

General models, specialist investment tools and internal agents each solve part of the problem. We want to understand the operating choices behind real adoption — including what teams decide not to use.

A five-layer research lens

We look beyond the model itself. At every layer, the question is the same: buy, build, or combine?

  1. 01

    Information foundation

    External data, internal notes, email and meeting records — and how teams judge source quality.

  2. 02

    Firm context

    CRM history, investment thesis, taxonomy, portfolio knowledge and prior decisions.

  3. 03

    Intelligence

    Screening, thesis fit, research, diligence and evidence traceability.

  4. 04

    Agents

    What can work continuously, what needs a specialist product and what teams build internally.

  5. 05

    Workflow and control

    Approvals, corrections, ownership, access and the human judgment teams retain.

The tools teams bring into the conversation

We are mapping how teams combine general AI, specialist intelligence products and their existing operating stack.

General AI

Foundation models and agentic assistants

  • Claude
  • ChatGPT
  • Gemini

Investment intelligence

Research, sourcing and private-market intelligence

  • Kruncher
  • Harmonic
  • Evertrace
  • Specter
  • Dealroom

Firm systems

Relationship context and the systems of record

  • Affinity
  • Attio
  • DealCloud

Company knowledge

Meeting intelligence, firm memory and shared context

  • Granola
  • Circleback
  • Notion
  • In-house / custom

Build and automation

Internal tools, agentic coding and workflow building

  • Claude Code
  • Codex
  • n8n

Always-on agents

Persistent agents that keep working across the operating cadence

  • OpenClaw
  • Hermes
  • AgentLed

Outreach

Prospecting, relationship activation and follow-up

  • Apollo
  • Clay
  • Instantly

Examples only — not a ranking, market-share claim or endorsement. The findings will report only de-identified patterns that emerge from the study.

Study response

What does your team use today?

Select the tools in your active stack, add anything we missed, and help us map the combinations investment teams are actually running.

We will use your domain only to identify the firm behind this response and send the de-identified findings brief when it is published.

Select every tool your team actively uses

General AI

Investment intelligence

Firm systems

Company knowledge

Build and automation

Always-on agents

Outreach

What best describes your in-house setup?
What technical capacity does the firm have? Select all that apply.
Which recent work triggered this decision? Select all that apply.
What mattered most in the decision? Select all that apply.
How would you like to take part?

Your response will be analysed only in de-identified aggregate patterns. We will never publish anything beyond an opt-in firm logo without separate approval.

We’ll email you the findings brief and practical patterns from peer investment teams when the study closes.

What participation involves

A 20-minute research conversation about one recent workflow, whether AI was used or not.

  • What triggered the work and which tools were involved
  • What a human reviewed, corrected or decided to keep manual
  • What changed — including no change, a paused experiment or a stopped tool
  • What evidence would make the next adoption decision easier
Share a workflow by email

Research boundaries

  • Not a sales call or product trial
  • No confidential data, credentials or system access requested
  • No recording or attribution without separate permission
  • Positive, neutral and negative experiences are equally useful

Participate in the study

Take part in the way that suits you. Everyone who opts in will receive the findings brief and practical industry insights.

01

Share your perspective

Share a 20-minute research perspective. Your contribution stays private and is used only in de-identified findings.

Share your perspective 02

Receive the findings only

Receive the de-identified findings brief when it is published, without taking part in an interview.

Share your perspective 03

Be considered for a featured contribution

Contribute to the study and signal that you may be open to a named contribution on this page.

Share your perspective

A public contribution is never automatic: we will request approval for the exact wording, name, role and placement before it appears on this page.

Share your perspective

A short research conversation, a findings brief, or a possible featured contribution — choose the participation that suits you.

Share your perspective

Back to VC Analyst