DavidAgents — DavidLake
What do we know?
Your agents build and run the data pipelines: scheduled workflows that pull from any system you use — SaaS APIs, spreadsheets, mailboxes, databases — do the data engineering, and land everything in your own lake: fast analytical storage backed by object storage that keeps history for pennies. Then you just ask questions.
The problem
Orders in one place, ad spend in another, payments in a third, the spreadsheet that actually runs things in a fourth. The industry's answer is a stack: a connector vendor, a transform tool, a warehouse, and an engineer to babysit all three — priced for companies with a data team. Small businesses get dashboards of one system at a time and gut feel for the rest. The gap isn't tooling. It's staffing. So we staffed it.
Here's what every page on this site is really about: this whole company — every product line on it — was built and is run by one founder and his agents, on the platform you're reading about. That team is the proof it works. What we sell is the same leverage, priced so anyone can have it — because how good you are should decide how far you go, not your software budget. Why we build this way →
Capabilities
The numbers behind the business
Every pull lands in your lake with its history, gets shaped into clean datasets and summaries, and comes out the other end as dashboards and feeds your team and your agents read from the same place. Refresh as often as the data deserves. New questions become queries, not projects.
● dashboards read straight from your lakeThe agent is the connector
Connector vendors sell you a menu; if your system isn't on it, you're stuck. Our agents don't have a menu — you hand over credentials through a secured vault form, the agent reads the API like an engineer would, and writes the pull itself. If it has an API, an export, or an inbox, it can join your lake.
● the agent writes each pull itself, from the source's API, export or inboxScheduled workflows
Every source becomes a scheduled job: nightly, hourly, whatever the data deserves. First run backfills history; every run after keeps the lake current. Each run journals its lineage — what ran, rows loaded, outcome, errors — into a pipelines panel you can actually read.
● every run leaves a receiptData engineering included
The reason pipelines rot is that sources drift: a field renames, a format changes, a pagination rule moves. Traditional stacks page an engineer. Here, consecutive failures — or a scheduled refresh that quietly stops — wake the agent that built the pipeline: it reads the error receipts, fixes the transform, and re-runs. And if it genuinely can't fix it, it doesn't retry forever in the dark: after three attempts you get one plain-language email, and the alerts pause until a run succeeds. Loads carry declared field types too, so a source that starts sending garbage gets rejected at the door — never silently averaged into your numbers.
● failures wake the engineer that never sleeps; humans hear about the ones it can't fixYour own lake
Underneath: a columnar analytics engine for fast questions, tiered onto object storage for history — the same architecture the big-data world runs, without the big-data invoice. Your lake is yours alone, walled off from every other business. Keep years of history; retention costs pennies, not seats.
● hot queries, cold penniesAsk it anything
Your manager agent queries the lake in plain conversation — "margin by product since June, weekly" — joining whatever datasets the question needs, and the answer lands as numbers, tables, and charts on your dashboard. Add a Data Analyst agent and reporting becomes someone's full-time job. The reports you already have get deeper the moment more of your data lands in one place.
● the dashboard you have, fed by everything you useGoverned like everything else
Source credentials go into the vault through a secure form — agents use them, never see them: the platform injects the key into the request and hands the agent back only data. And a credential can only ever travel to a host you've approved, over TLS, with redirects refused — the same governance discipline that gates our agents' calls and email gates your data's movement.
● creds in the vault, every pull to an owner-approved hostNo surprise bills, by construction
Warehouses bill by the query and tell you at the end of the month. Here every query returns a cost receipt — rows scanned, bytes read, milliseconds — and queries that would scan wastefully are refused with advice, not billed. Your agent designs reports against those receipts, pre-aggregates the heavy ones, and the flat price stays flat. The lake literally will not let itself become expensive.
● wasteful queries get rejected with a lesson, not a line itemWhat's inside
How it lands
"Pull my Shopify orders and my ad spend, nightly." The agent asks for credentials through a vault form and proposes the pipeline.
It writes the pull, shapes the tables, loads the history, and schedules the cadence — journaling every run.
Questions in chat become charts on your dashboard. New sources deepen every answer you already get.
Pricing
Every AgentsFast subscription includes a working lake — 2GB stored, pipelines, dashboards, the lot. Storage packages expand it flat: never per-seat, never per-query, and the caps are deliberately generous because your data warehouse should not be a second rent.
$29/mo
Room for years of orders, deals, bookings and campaign history. One sentence in chat turns it on.
$79/mo
Serious multi-source history with the same flat bill. Compare that to any warehouse invoice.
$199/mo
Half a terabyte, engineered and charted by agents. Beyond that: DavidAgents, your own deployment.
Proof
DavidLake comes from the team that runs Cortex — our platform's own data plane — built the same way and opened to your business's data: agents building the pipelines, a columnar engine answering, cheap storage keeping the history.
Same platform we run our own company on. First line is a purchase — the second is a decision.
Bring my data together