SB StartupBasket
All ideas
74 /100 GO Medium complexity

DriftLog — fuel-log data sentry for small ship managers

Catches bad fuel-consumption logs while the voyage is still fresh, not in 2027 when the verifier withholds the certificate.

Logistics SaaS Global Compliance-drivenSMBMaritimeAI-agentSolo-builder
— views
Evaluation Scores
74/100

GO

Overall Score

16
Problem
11
Demand
11
Build
11
Distrib.
12
Revenue
7
Time
6
Defense

DriftLog

1. One-liner

Catches bad fuel-consumption logs while the voyage is still fresh, not in 2027 when the verifier withholds the certificate.

2. Trend signal — why now?

On 1 January 2026 the rules for what ships must write down changed, and almost nobody re-tooled for it.

Every ship of 5,000 GT and above on international voyages now has to collect IMO DCS data at a new level of granularity: fuel consumption broken out by consumer group — main engine, auxiliary engines, boilers — split into underway and not underway, with BOSP/EOSP event marking, shore power reporting, and port call purpose classification. That’s the MEPC 81 amendment package, in force 1 August 2025, with the first full calendar year of enhanced collection running right now.

Here’s the part that makes this a business. Collection and reporting are separated in time. The first enhanced reports don’t get submitted until 2027 — data collected during calendar 2026 must go for verification no later than 31 March 2027. So every logging error being made today sits undiscovered for up to fifteen months, and by the time a verifier asks about it, the voyage is long over and there is nothing left to re-measure.

The industry’s own vendors are saying this out loud. Uwe Krüger, Joint Managing Director at CMT, frames expanded DCS as “a practical shift from estimation and reconciliation towards direct measurement,” and warns that operators wrongly assume this can be handled with revised reporting procedures rather than changes to how data is generated onboard. His words on the stakes: “the quality of those submissions depends entirely on what is measured and logged onboard throughout 2026.” When the data doesn’t hold up, “incomplete or inconsistent datasets can result in a SoC being withheld.”

A withheld Statement of Compliance is not a paperwork slap. A missing or expired SoC is a deficiency inspectors can cite, and in repeat cases use as grounds for detention.

Meanwhile the buyer I care about is running on Excel. Small fleet operators “rely on spreadsheets, paper logbooks, and email”; their data “exists, but it’s scattered across noon reports, emails, Excel, bunker delivery notes, and different vendor portals, making compliance and DCS reporting difficult.”

And the regulatory pressure isn’t easing. The IMO Net-Zero Framework adoption was adjourned in October 2025 by a 57–49 vote and pushed to October 2026, delaying a global framework to roughly 2028 — which, per DNV, increases “the risk of a fragmented regulatory landscape in the meantime.” Fragmented means owners keep running IMO DCS and EU MRV as two overlapping datasets with two separate penalty regimes. More reconciliation, not less.

Scope check: for reporting year 2023, 28,620 ships across 105 flag administrations submitted DCS data. This is a defined, enumerable, regulated population.

Provenance:

3. The opportunity

The gap is a timing gap, and timing gaps are the best kind because they close on a deadline whether the customer acts or not.

Enhanced DCS moved the work from reporting to logging. The old game was: collect aggregate fuel figures all year, reconcile them at the end, submit. You could paper over a sloppy month during compilation because the required output was coarse. The new game demands per-consumer, per-state granularity that has to be right at the moment the crew writes it down. You cannot reconstruct in March 2027 how much fuel the auxiliary engines burned while not underway on a Tuesday in June 2026. The measurement either happened or it didn’t.

So there’s a fifteen-month window where a fleet is quietly accumulating a defective dataset, feels completely fine, and has no signal that anything is wrong.

Who’s serving this? At the top, ZeroNorth and StormGeo — bespoke, enterprise-grade, custom onboarding, fleet-size-dependent contracts. Class societies (DNV, ABS, Bureau Veritas) offer DCS services, but they’re the verifier or adjacent to it; they tell you the answer at verification time, which is exactly too late. At the small end, Navatom and NOZZLE sell general ship-management suites where DCS is one module among crew, maintenance, and certificates — broad and shallow, built around the noon report as a record, not as something to interrogate.

Nobody in that lineup is doing the narrow, obsessive thing: reading every noon report the day it lands, cross-checking it against the bunker delivery notes and the previous report, and telling a superintendent this week that vessel three has been logging boiler consumption into the aux engine bucket since April.

That’s the 10× move. Not a better dashboard — a shorter feedback loop. The incumbent tells you at verification; this tells you while the fix is still free.

4. Target market

  • Primary customer: Technical superintendent or DPA at a ship-management company or owner-operator running 3–25 vessels of 5,000 GT and above on international voyages. Greece, Turkey, Singapore, Cyprus, UAE, Japan, Norway. Often the DPA works from home or manages the fleet alongside other duties — this is a small back office, not a corporate HSEQ department.
  • Why they buy: Because their data “exists, but it’s scattered across noon reports, emails, Excel, bunker delivery notes, and different vendor portals.” They know 2026 logging is different. They do not know whether the last seven months of it are defensible, and they have no cheap way to find out before the verifier tells them.
  • Rough TAM reasoning: 28,620 ships submitted DCS data for reporting year 2023 across 105 flags. The enterprise platforms have the top of that. If the 3–25 vessel segment is even a quarter of the fleet, that’s ~7,000 vessels reachable, spread across a few thousand managers. At per-vessel pricing, a few hundred vessels is a real business.
  • Why now for them: The 2026 calendar year is the data. It’s half-spent. Every month they wait is a month of logs that can’t be re-measured, and the submission deadline of 31 March 2027 is fixed.

5. Product sketch (MVP)

  • Ingest what already exists — noon reports (email, Excel, PDF), bunker delivery notes, and existing DCS spreadsheets. No onboard hardware, no new crew workflow on day one.
  • Enhanced-granularity gap check — flags where a vessel is not reporting fuel by consumer group, missing the underway/not-underway split, missing BOSP/EOSP event marking, shore power, or port call purpose. The literal 2026 checklist.
  • Cross-source reconciliation — noon-report consumption vs bunker delivery notes vs voyage records, with the discrepancies ranked by how much they’d matter to a verifier.
  • Drift detection — catches a vessel whose logging pattern silently changed (new chief engineer, new template, a consumer bucket that went to zero and stayed there).
  • Weekly superintendent digest — one email per fleet: what’s wrong, which vessel, which voyage, what to tell the master this week.
  • Verification dry-run — generates the dataset as a verifier would see it, with an honest list of what will draw a clarification request.
  • Dual-regime view — the same underlying voyage data shown against IMO DCS and EU MRV side by side, so the two submissions don’t contradict each other.
  • Fix log — a timestamped record of every issue raised and what was done, which is the thing you hand a verifier who asks why June looks different from May.

6. AI angle — what’s load-bearing

Take the AI out and this doesn’t work, because the input is unstructured and inconsistent by nature.

Noon reports arrive as email bodies, Excel attachments with a different layout per vessel, scanned PDFs, and occasionally a photo of a printed form. Bunker delivery notes are scans. The same fleet will have four template dialects because four different technical managers set them up across a decade. Building deterministic parsers for that is whack-a-mole and it’s precisely why small operators gave up and left the data scattered.

Document extraction models handle this now at a cost that makes per-vessel pricing work. That’s the load-bearing part: turning heterogeneous vessel paperwork into a comparable time series without asking the crew to change how they report.

The second AI job is anomaly reasoning. “Aux engine consumption while not underway dropped 60% after 12 April and never recovered” is not a threshold rule — it needs the model to weigh the vessel’s own history, its sister ships, and whether a port call or a crew change explains it. Flagging every deviation would be noise, and noise is what gets a compliance tool deleted.

What AI must not do here: assert a number. The product never invents a consumption figure to fill a gap. It says “this is missing and cannot be reconstructed” — because fabricating data is exactly the failure mode that gets an SoC withheld.

7. Localization angle (if any)

N/A — this is a global play. IMO DCS is a flag-state regime applied identically to ships on international voyages regardless of owner nationality, so the compliance logic is uniform worldwide. The only regional variation worth building is the EU MRV overlay for EEA port calls, which is a second ruleset rather than a localization.

There is a distribution geography, though, and it’s worth naming: small and mid-size ownership concentrates heavily in Greece, Turkey, Cyprus, Singapore and the UAE. Piraeus in particular is a dense, walkable, conference-driven market where a few hundred managers control thousands of vessels. That’s a sales geography, not a product one. English is the working language of the industry; no multilingual build needed for v1.

8. Business model — path to $1M–$5M ARR

  • Pricing: $180/vessel/month, billed annually per fleet. Sits deliberately below enterprise platforms (StormGeo benchmarked at roughly $35–$135/unit/month for fleet management, ZeroNorth bespoke/custom) while being a specialist tool rather than a suite. A 10-vessel manager pays $21,600/year — small against the cost of one withheld SoC or one detention.
  • ACV: ~$21,600 at 10 vessels; realistic blended ACV of $18,000–$25,000 given the 3–25 vessel target band.
  • Rough math to $1M ARR: 48 fleets averaging 10 vessels = 480 vessels × $180 × 12 = $1.04M. Forty-eight customers. That is a callable list, not a marketing funnel.
  • Rough math to $5M ARR: ~2,300 vessels — roughly 230 fleets, or about 8% of the 28,620-ship DCS population. Requires expanding upward into 25–60 vessel managers and adding the EU MRV/FuelEU workload as a second paid module.
  • Expansion path: More vessels per fleet is the natural motion (fleets grow, and land-and-expand from a 3-ship pilot to the whole fleet is the standard maritime software path). Then per-regime modules as the framework lands — if the IMO Net-Zero Framework is adopted in October 2026, every owner needs a third dataset reconciled against the first two, and that’s an upsell that writes itself.

9. Go-to-market wedge — first 100 customers

  • Sell the audit, not the software. Offer a free “2026 DCS readiness check”: send us six months of noon reports for one vessel, get back a written list of what a verifier will question. This is deliverable in a day, it’s genuinely alarming when the data is bad, and it converts because the owner cannot self-diagnose. This is the wedge — everything else is how you get the reports.
  • Work the verifier and consultant channel. Independent DCS/MRV verifiers (Normec Verifavia and peers) and technical consultancies see the bad data first and are structurally forbidden from fixing it for the clients they verify. They have every incentive to refer a remediation tool. Ten referral relationships is a pipeline.
  • Piraeus and Singapore, in person. Small-owner maritime buys through relationships and conferences, not inbound. A two-week trip to Piraeus with 30 pre-booked meetings, running the free readiness check live on a laptop, is a realistic path to the first 10 paying fleets. Repeat in Singapore and Istanbul.
  • Flag-state and class bulletins as a hook. Owners actively read circulars from their flag administration and class society about enhanced DCS. Publishing a plain-English “what changed on 1 January 2026 and how to check your own logs” guide, timed against those bulletins, is the one content play that works here because the audience is already searching for it.
  • Target the fleets that just changed something. Change of flag, change of manager, or a new technical superintendent all reset logging conventions mid-year — the exact condition that produces drift. These are trackable events and they make for a specific, well-timed cold email.

10. Build complexity — justification

Medium. The parsing layer is the real work: noon reports and bunker delivery notes in inconsistent formats across vessels and managers, which is document-extraction work plus a lot of unglamorous per-fleet template tuning in the first months. Off-the-shelf: the extraction models, the web stack, email ingestion. Custom: the enhanced-DCS ruleset (consumer-group mapping, underway/not-underway logic, BOSP/EOSP handling), the reconciliation engine, and the EU MRV crosswalk. No onboard hardware, no vessel installation, no integration with engine systems for v1 — that’s what keeps this out of High. Call it 12–16 weeks to a v1 a design partner can run on a live fleet, with domain review from someone who has actually filed a DCS submission.

11. Gating checklist

GatePass?Note
Legal in target market✅Helps owners meet an existing IMO obligation. Product never fabricates data — it flags gaps as unreconstructable.
Ethical — no harm / dark patterns✅Improves the integrity of emissions data. The refusal to invent numbers is a design principle, not a limitation.
Market exists (evidence above)✅28,620 ships in DCS scope; enhanced granularity mandatory from 1 Jan 2026; verification deadline 31 Mar 2027.
1–5 person team can build this✅12–16 weeks with a technical founder plus a maritime domain advisor.
Launchable with <$50K / ₹40L✅Software-only. Main costs are extraction API spend and a Piraeus/Singapore sales trip.

12. Feasibility score

AxisWeightScoreNotes
Problem intensity2016/20Real and consequential — withheld SoC, citable deficiency, detention grounds in repeat cases. Docked because the pain is latent: it doesn’t hurt until 2027, and latent pain is harder to sell than bleeding pain.
Demand evidence1511/15Regulation is hard-dated and the scope is enumerable; industry vendors publicly flagging the gap; small operators documented as spreadsheet-bound. But I have no verbatim owner complaints — maritime pain lives in private channels and conferences, not public forums. That’s a real evidence gap and I’m scoring it honestly.
Build feasibility1511/15Software-only, off-the-shelf extraction, no hardware. Docked for messy multi-format parsing and per-fleet template tuning.
Distribution clarity1511/15Free readiness check is a strong, concrete wedge; verifier/consultant referrals are a real channel; enumerable target list. Docked because maritime is relationship-driven and requires travel — not a 2-week sprint.
Revenue mechanics1512/15Per-vessel pricing is the category norm, benchmarked against known platforms; 48 customers to $1M is very achievable. Docked because small owners are famously tight and unproven at this price point.
Time to first revenue107/10Design partner can pay within 8–10 weeks off the readiness check; full sales cycle in maritime runs longer than SaaS norms.
Defensibility106/10Soft moat: accumulated per-fleet parsing templates, a growing library of what actually triggers verifier queries, and workflow lock-in once the fix log becomes the audit trail. But the ruleset is public and an incumbent could bolt this on.
Total10074/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required

This needs someone who can build a document pipeline and a co-founder or advisor who has personally filed a DCS submission or worked as a verifier. Without the domain half, the rules get subtly wrong and the product loses credibility on the first call. Don’t attempt this as a pure outsider.

Key assumptions to validate (3–5)

  1. Assumption: A meaningful share of 3–25 vessel fleets actually have defective enhanced-DCS logging in their 2026 data right now. How to test: Run the free readiness check on 15 vessels across 8 different managers. If fewer than half show material gaps, the premise is wrong and this is a much smaller business.
  2. Assumption: Superintendents will pay ~$180/vessel/month for a specialist checker on top of whatever suite they already have. How to test: Price conversations with 20 managers after showing them their own bad data. Look for willingness to sign a pilot, not politeness.
  3. Assumption: Verifiers and consultants will refer rather than build this themselves. How to test: Direct conversations with 5–8 independent verification firms about a referral arrangement.
  4. Assumption: Noon reports and BDNs can be parsed reliably enough across template dialects that per-fleet setup stays under a few hours. How to test: Collect real report samples from 10 different managers and measure extraction accuracy and tuning time before writing the product.

Risk flags

  1. Latent-pain risk: This is the big one. The consequence lands in 2027; the spend is required in 2026. Compliance products that ask for money before the pain arrives have a harder sale than the regulation implies. The free readiness check exists specifically to make latent pain visible, and if it doesn’t do that, the whole GTM collapses.
  2. Incumbent absorption: ZeroNorth, StormGeo, DNV, or a suite like Navatom could add enhanced-DCS validation as a feature. The ruleset is public. The defense is being narrower and faster than a suite wants to be, and owning the verifier referral relationships early.
  3. Regulatory timing: The IMO Net-Zero Framework was adjourned to October 2026. If it’s adopted, the compliance workload grows and this gets more valuable. If it collapses further, some owners may read the whole direction of travel as softening and defer spend — even though DCS itself is unaffected.
  4. Market access: Small maritime is relationship-gated and travel-dependent. A founder who can’t get to Piraeus, Singapore or Istanbul will struggle to reach the buyer regardless of product quality.

14. Structured verdict

Score:                  74/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder who can build a document-extraction pipeline, paired with a
                        maritime domain co-founder or advisor with DCS filing or verification experience
Time to revenue:        8–12 weeks to first paid design partner
Capital to launch:      $15–25K (extraction API spend, domain advisor, one Piraeus/Singapore sales trip)
Top 3 assumptions to validate first:
  1. Real fleets have defective 2026 enhanced-DCS logs — run free readiness checks on 15 vessels
     across 8 managers, measure the share with material gaps
  2. Superintendents pay ~$180/vessel/month for a specialist checker — 20 pricing conversations
     held immediately after showing them their own data
  3. Verifiers and consultants will refer rather than compete — direct talks with 5–8 firms
Kill criteria:
  - Abandon if fewer than 50% of audited vessels show material enhanced-DCS logging gaps
  - Abandon if under 3 of 20 managers will start a paid pilot after seeing their own defective data
  - Abandon if an incumbent suite ships equivalent enhanced-DCS validation before your v1 and
    bundles it at no additional cost

15. Next step — 1-week validation sprint

  • Day 1–2: Build the target list. Identify 40 ship managers in the 3–25 vessel band across Greece, Turkey, Cyprus and Singapore from flag-state registries and class records. Simultaneously, get the enhanced-DCS ruleset written down properly with a domain advisor — the exact fields required from 1 Jan 2026 and what a verifier queries.
  • Day 3–4: Offer the free readiness check to all 40. Target 10 managers sending six months of noon reports for at least one vessel. Analyze by hand — no product, no automation. The point is to find out whether the data is actually broken.
  • Day 5: Deliver findings live to every manager who participated and immediately ask for a paid pilot at $180/vessel/month.

Falsifiable outcome: Of the vessels audited, what percentage show material enhanced-DCS gaps, and how many managers commit to a paid pilot on the call? Proceed only if ≥50% of vessels show material gaps AND ≥3 of 20 managers commit to a pilot. Below either threshold, the latent-pain risk is confirmed fatal and I walk away rather than spend six months building for a deadline the market intends to ignore.

Interested in a detailed proposal?

Get a deep-dive with market research, competitive analysis, and implementation roadmap.

Contact us

info@startupbasket.ai