Sediment 1
Know which work is worth your time and money.
17x cheaper and 2x faster than frontier models.
Your systems record every job, deal, and project. Nothing reads across them. Sediment 1 does, and tells you where your estimates miss, which rates to fix, and which work is worth taking. ChatGPT, Claude, and Glean only see the few files a search turns up. Sediment 1 is trained on all of them.
Where you make the most money
It does not wait to be asked.
Sediment 1 reads everything your company produces as it arrives and tells the right person what matters before anyone thinks to ask.
Trained on everything
Not a search over a few files. Every record you have.
Works on its own
Flags problems and openings as new work arrives.
Keeps learning
Every outcome sharpens the next answer.
Shows its source
Every answer links to the records behind it.
A view for every role
Owners, operators, finance, and sales each see what matters to them.
Built around how you work.
Most tools make you work their way. We build Sediment 1 around yours. Every company gets its own interface, built around its roles and its numbers.
Bids
What changedJ2672 closed 17% under on labor, so I moved the Vaultline bid down 25 hours.
Open bids
Tres Rios Gilbert Medical Center 4th Floor Patient Wing
Calloway Brandt Construction · Hospital · Renovation (occupied)
Your estimate looks 477 hours low (+$29,831).
This GC has back-charged you before
Vaultline CHD1 Data Hall 4 Mechanical Fit-Out
Corrigan Voss Builders · Data center · New construction
Your estimate looks 633 hours high.
Aurelis Tempe Fab Support Building Cleanroom Bay 4
Granite Reef Constructors · Manufacturing · Renovation (occupied)
Not enough similar past jobs for 57% of this bid.
Starts from your exports
No cleanup. We read your files as they are.
Unlimited usage
We have a flat price, so you can ask as many questions as your team wants.
Your own machine
Dedicated to your company, with no calls to outside providers.
See how it works
| Attribute | Sediment 1 | Opus 5 | GPT-5.6 Sol | Claude Code |
|---|---|---|---|---|
| Mean cost per query | $0.01 | $0.25 | $0.09 | $0.48 |
| Median speed per query | 20.1s | 40.5s | 51.8s | 24.8s |
Accuracy by task family
- Sediment 1
- Opus 5
- GPT-5.6 Sol
Same model, four setups
Sediment 1 starts from an open-weights model that runs on a machine dedicated to you. On top of it we add two things. The first is training on your company's records, so the model already knows your history before the first question. The second is our own retrieval and orchestration harness, which finds the right records for each question and attaches the source to every answer.
To test it, we built two synthetic institutions, a private equity firm and an investment bank, each with roughly 25 million tokens of internal documents. We asked five kinds of question. Recall a fact, use the newest version of it, decline when the answer is not there, apply rules nobody wrote down, and draft in the house format. We ran Opus 5 and GPT-5.6 Sol inside the same harness, so the only difference was the model, and we ran Claude Code with Opus 5 on its own over the same documents. The benchmark harness is public on GitHub. The documents and answer keys stay private, which keeps the benchmark valid.