AI and software for contractors, concessionaires, developers and awarding authorities, across claims, project controls, procurement and assets.
The gap is almost never the entitlement. It is the record that would have proved it.
In brief
EigenSpark builds AI systems and the software around them for EPC contractors, concessionaires, developers and awarding authorities. Your sites already generate daily progress reports, measurement books, correspondence, drawings and photographs. We read all of it, tie each event to the contract clause it triggers, and keep the register current while the project runs. The same records then drive progress, procurement and asset work.
What we cover
Model agreements, dispute rules, clearances and site compliance.
The starting position
Thin evidence behind real entitlements, unreadable daily reports, and progress nobody can prove.
A verbal instruction, an undated photograph, a daily report that describes only the weather. The gap shows up years after it could have been closed.
Every site produces one and almost nobody can search them. The deployment record and the hindrance log are buried inside.
The programme, the measurement book and the bill disagree. Every forecast of cost to complete inherits that disagreement.
Notice periods, certification windows and escalation formulae, in a document nobody opens when it matters. A missed notice kills a good claim.
How an engagement runs
Read the records you already produce, set up one project, then carry it to the rest.
Two to three weeks inside your daily progress reports, measurement books, correspondence, programme and billing data for one project, to establish what is capturable today and where the evidence gaps sit on live entitlements.
A single project instrumented: the obligation register live against the programme, daily records structured as they arrive, and one claim assembled from them. Small enough to prove in a quarter, and it pays for itself on one entitlement.
Ingestion, the obligation library, the record structure and the claim templates are shared. The second project is set-up, and your contracts team maintains the library.
Use cases
Claims and contracts, then project controls, then procurement and assets, then the back office.
The entitlement, the clause, the chronology and the records assembled into one pack, with every assertion carrying the report or letter it rests on.
Continuous Compliance Monitor →Notice periods, milestones, certification windows and escalation formulae extracted into a live register, so an obligation is flagged while it can still be met.
Contract Risk Analyser →The as-planned and as-built programme compared activity by activity, with hindrances from the daily record mapped on, so the delay story is built from evidence.
AI & Data Strategy →Site reports read into structured records: activities, deployment, plant hours, weather and instructions, so a question about last October takes a minute.
Document Intelligence Engine →Committed cost, measured work, certified amounts and the remaining programme read into one forecast, and every disagreement between the three surfaced by name.
Data Engineering & Platforms →Deployment, output, rework, safety and billing read together per subcontractor and per package, so a package heading for trouble shows up early.
AI & Data Strategy →Bids read against the eligibility criteria, the priced bill of quantities and the specification, with every deviation linked to the clause it fails.
Contract Risk Analyser →Purchase, receipt, consumption and theoretical requirement reconciled per material and per structure, with the variance ranked by value.
Data Engineering & Platforms →Log sheets, fuel issues and deployment records read into utilisation and idling per machine, which supports both the hire decision and the idling claim.
Predictive Maintenance System →Consent conditions, monitoring results and labour registers assembled continuously against the obligations that apply, with lapses flagged early enough to fix.
Continuous Compliance Monitor →Incidents, near misses and toolbox records read for the conditions that keep repeating, with the statutory reporting clock tracked against each event.
AI & Data Strategy →Drawing revisions, requests for information and design changes tracked against the activities they affect, with the cost and time consequence recorded on arrival.
Document Intelligence Engine →The engineering half
Four layers of ordinary engineering. The first one carries every claim you make.
Daily reports, measurement books and correspondence exist on every project. Almost none of it is queryable, and that is the whole opportunity.
The ERP and the scheduling tool stay. Most of the effort is connecting them to what the site actually records.
Built for a site engineer on a phone with no signal and a contracts manager assembling a pack, with the source of every assertion one click away.
Defined once and used everywhere, so the progress review, the cost report and the claim cannot disagree. Your team runs and owns it.
We also run the training that goes with it: SQL and data engineering, cloud, enterprise systems and cybersecurity, alongside the AI tracks. A team that cannot query its own data cannot specify a model either.
What we offer infrastructure and public works teams
Most teams start with one project and the contract that governs it.
Where most project engagements start.
The build, and the engineering under it.
AI tracks, and the foundations under them.
The teams that own this work.
Working in a specific function? See how we help Project Management teams.
Security and controls
The claim stays yours. Every assertion links to a dated record that already existed.
We build the register, the record and the pack. The entitlement position, the legal strategy and the submission stay with your contracts team and your counsel.
A statement in a claim pack carries the daily report, the letter or the measurement it rests on, with the date it was created. A record assembled afterwards is worth very little, and the system is built to prevent that.
Field capture holds its state offline and synchronises when connectivity returns, because the records that matter most are made at the places with the worst network.
Clauses, notice periods, milestone definitions and escalation formulae are managed by your contracts team. A new form of contract is configuration.
FAQs
NHAI's January 2026 circular splits the dispute route by value on its model agreements: claims below Rs 10 crore go to arbitration and larger ones are pushed to conciliation or the civil courts. A conciliation room and a court timeline both reward the party that already holds dated, contemporaneous records, and neither will order the discovery a tribunal might have. The practical answer is to capture the record while the work is happening, which is what the first two tracks on this page do.
We build the register, the structured record and the assembled pack, with every assertion linked to the document it rests on and the evidence gaps listed. The entitlement position, the commercial strategy and the submission itself stay with your contracts team and your counsel. We are engineers, and we are clear about where that ends.
Yes, and they are usually the single most valuable asset on the project. Site reports are read into structured records covering activities, deployment, plant hours, weather, hindrances and instructions, which makes the whole history queryable. Delay analysis, idling claims and subcontractor performance all come out of the same structured record.
It is built for that. Field capture holds its state offline and synchronises when connectivity returns, because the records that matter most are made at the places with the worst network. Nothing that a site engineer has to do depends on a live connection.
Yes, and mostly the same components read from the other side. Bid scrutiny, the obligation register, progress and cost verification, material reconciliation and statutory compliance evidence all apply. Our public sector page covers the governance and audit requirements that come with being the awarding body.
Usually on the first entitlement it saves. The obligation register alone catches notice periods that would otherwise close, and a single missed notice on a large project is worth more than the engagement. The structured record work takes a quarter on the first project and materially less on the second.
No, and the parts that are not are usually what makes the AI work. Integration with the ERP and the scheduling tool, the record structure, the field application and the obligation library are ordinary engineering, and they are most of the effort. The same is true of the training: alongside the AI tracks we run SQL, data engineering, cloud, enterprise systems and cybersecurity.
In your own tenancy. Project records are commercially sensitive and frequently disputed, so they stay inside your boundary with access logged. Where a hosted model is used for text work, what is sent to it is agreed and recorded before build.
Related
The two products this ships as, the service under them, and the teams that own it.
Obligations, variations and bid comparison.
→ ProductClaim packs, statutory evidence and inspections.
→ ServiceSite records, programme and billing in one place.
→ FunctionThe function that owns programme and cost control.
→ IndustryThe authority side, and the governance around it.
→ PillarThe engineering side, for systems that run on site.
→One entitlement you are pursuing, with the daily reports and correspondence for the period. We call you within 48 hours and go through what the record already supports, where the evidence gaps are, and what it would take to close them.
Talk to us