Before touching a single tool, we ran deep discovery sessions with every member of the team — sitting in on live work, shadowing calls, and mapping the unspoken logic behind how decisions actually got made.
Manually triaged hundreds of inbound submissions, with no systematic way to surface deals aligned to niche theses like sugar reduction or real-world decentralization. High-potential companies were being missed simply because there weren't enough hours in the day.
Reviewed decks entirely by hand, wrote rejection emails manually — often weeks after submission. No structured framework existed for evaluating founder quality across calls. Institutional knowledge lived only in his head.
The entire KYB/KYC compliance process sat on her shoulders. Founders drip-sending documents, manual Drive organization, manual Carta uploads — bad file naming alone was adding 30–40% overhead to every single deal close.
Maintained qualitative portfolio data in Airtable with no connection to Tactic (financial data) or Attio (CRM). Valuable context about portfolio companies existed only in her spreadsheet, invisible to the rest of the team.
The real insight wasn't any single broken process. It was that each person had built their own private system — and none of them talked to each other.
Observation from Discovery · Tetris LabsDiscovery surfaced one structural truth: the firm had four highly capable people operating as four separate firms. Data didn't flow. Decisions couldn't compound on prior knowledge. And every hour spent on manual ops was an hour not spent on the work that actually moves the needle — finding and backing the next breakthrough company.
The output was a complete operational map across four functional areas: Deal Sourcing, Deal Evaluation, Investment Closing, and Portfolio Management — with clear friction points, tool dependencies, and automation opportunity for each.
The problems weren't technical — they were structural. Manual processes had quietly become the load-bearing walls of the firm's operations. Everyone had adapted to the inefficiency rather than challenged it.
Deal flow was 100% inbound. No system existed to proactively find companies in the narrow spaces the fund genuinely cared about. Edge bets in niche theses — the exact areas where the fund had conviction — were being systematically missed.
Decks sat unreviewed. Rejections took weeks. No structured scoring framework existed for founder calls. Institutional knowledge — memos, call notes, LP feedback — was scattered across six tools with no way to query or compound it.
KYB/KYC was fully manual — email chains, drag-and-drop downloads, individual Carta uploads. Founders sent documents in multiple tranches with inconsistent naming. No checklist. No visibility. Every close carried significant avoidable overhead.
Financial data lived in Tactic (accessible to only one team member). Qualitative data lived in Airtable (maintained by just one other). CRM data lived in Attio. No integration between them. Portfolio visibility required assembling three separate sources manually.
Every solution was designed to slot into existing behavioural patterns — not replace them. The team shouldn't have to learn a new way of working. The infrastructure should simply make the way they already work dramatically faster and more intelligent.
We didn't redesign their workflows. We automated the version of their workflow that existed in their heads — the ideal version they never had time to execute manually.
Thesis-driven sourcing agent that crawls niche data sources — forums, newsletters, Discord communities, Product Hunt, Hacker News — across the fund's conviction areas (e.g., real-world decentralization, sugar reduction, alternative proteins). Each company is enriched via Crunchbase/Clay, scored against a weighted rubric (founder background, traction, community sentiment, market fit), and the top deals each week are auto-posted to Slack with a full breakdown card.
Automated ingestion of funding announcements from newsletters and deal sources directly into Airtable, with automatic creation of the relevant pipeline records in Attio. No more manually tracking rounds from email.
A fleet of market intel scouts deployed via Yutori, tracking the entire web across the fund's thesis areas — funding announcements, partnerships, research, regulatory developments. Stories are published instantly to an internal news portal, tagged by category. Every Friday, a curated digest is assembled automatically and delivered to the relevant team members.
Captured all links from the team's #reading Slack channel into a structured Airtable base. Each link is auto-described and tagged by category. Every week, the agent assembles a Google Doc newsletter using all links from that week, structured using the team's established newsletter format and trained on all previous editions.
AI-powered analysis of founder call recordings. Transcripts are segmented into phases (intro, traction, team, close) and scored against a structured evaluation matrix across dimensions like market knowledge, communication clarity, and confidence. Scores are logged automatically into the deal record in Attio alongside reasoning. Designed to provide 30–50% more data-driven signal per founder evaluation.
Deals untagged as "Review" or "Consider" after 48 hours trigger an LLM-personalized rejection email — referencing the founder's name and company — pending one-click Slack approval from the team. Eliminates the founder limbo problem entirely. The partners maintain full oversight; the automation eliminates the friction that caused weeks-long delays.
All of the fund's institutional knowledge — deal notes, investment memos, call recordings, LP feedback — indexed into a vector store and made queryable via conversational AI. The team can now ask "Show all hydrogen-related startups we vetted in Q1 2025 with ARR > $100K" or "What did LPs say about our robotics investments?" and get an answer in seconds rather than hours.
Investment memos auto-generated from deal data, call notes, and enriched company information already captured in Attio and Google Drive. First-pass memos produced in minutes rather than hours, ready for partner review and refinement.
Gmail threads labelled with a custom KYB tag are watched automatically. Attachments are downloaded, intelligently renamed using LLM classification (pattern: Company_DocType_Date.pdf), and moved to the canonical Drive path — eliminating the 30–40% overhead that came from manual renaming alone. A KYB Tracker auto-populates with document status (Missing / Received / Uploaded), with Slack reminders for outstanding items.
Magic-link portal for founders to upload KYB/KYC documents directly. Each founder sees a personalized checklist rendered from Airtable. Files are auto-organized in Drive, doc types detected automatically, and progress is visible to both sides. Eliminates the multi-batch email chaos that previously defined every deal close.
When a deal moves to "Invested" status in Attio, a structured Google Drive folder is automatically created for the portfolio company and all existing deal files are migrated in. Teams continue using Drive as normal — the system just ensures structure appears at exactly the right moment.
Two-way sync connecting financial data (Tactyc: valuations, ownership percentages, KPIs), qualitative data (Airtable: runway, cash burn), and CRM data (Attio). A shared company ID as the primary key. Daily sync schedule. Simplified Attio views now give the entire team real-time visibility into key financial metrics that previously only one team member could access.
Investment documents (SAFEs, term sheets, KYC files) automatically synced from Google Drive to Tactyc's data room. Documents tagged with metadata on upload. Eliminates the redundant manual upload process previously run for every investment.
A complete automated attendee journey: online application → instant Attio sync and auto-enrichment → approval triggers a custom Stripe payment link sent via personalized LLM-generated email → 3-email follow-up sequence using enriched attendee data → on payment, automatically marked as confirmed in CRM, added to the Luma event, and sent a confirmation. Zero manual steps between application and arrival.
The goal was never to automate for automation's sake. It was to give a small, ambitious team the operational leverage to operate at the level their conviction deserved.
For the first time, the entire team has a unified view of every portfolio company — financial data, qualitative metrics, and deal history, accessible in one place without assembling it manually from three tools.
Thesis-driven deal sourcing is now proactive, not passive — the firm can now systematically see every company operating in the narrow spaces it cares most about, rather than relying on what shows up in the inbox.
Institutional knowledge that lived scattered across six tools is now queryable in seconds — memos, call notes, LP feedback, and deal history accessible through a single conversational interface.
The team's market intelligence now arrives in real-time and in summary — a continuous news portal and weekly digest replacing the ad-hoc tab-checking and Slack link sharing that passed for staying informed.