One point of fact worth getting right, because it appears incorrectly in a great deal of procurement documentation: the i2 portfolio is no longer an IBM product. IBM divested the entire i2 line to N. Harris Computer Corporation, part of Constellation Software, effective January 2022, and the products are now offered by i2 Group. Tender documents still referring to “IBM i2” are describing a commercial relationship that ended some years ago.
What follows is an evaluation framework rather than a comparison. It is written to help a unit decide what it actually needs before it looks at products.
First, establish why you are evaluating
Alternatives are usually assessed for one of five reasons, and each points toward different criteria.
Cost at scale. Desktop-licensed analytical tools become expensive as analyst headcount grows. If this is the driver, model total cost at your projected three-year headcount, including training and support, not at current seat count.
Data sovereignty and deployment control. If the requirement is that case data never leaves national infrastructure or a specific network, this becomes a threshold criterion that eliminates most options before functionality is considered.
Ingestion of local data formats. International products are generally strong on analysis and weaker on ingesting Indian call record formats, bank statement layouts, regional-language documents and scanned material. If your analysts spend most of their time preparing data, ingestion is your real requirement.
Scale of data. Desktop chart-centric tools handle case-sized datasets well and organisation-sized datasets poorly. If you need to query across all cases rather than within one, you need a server-side data platform, which is a different architecture.
Collaboration. Chart files on individual machines do not support multiple analysts working a common dataset with shared entity resolution.
Be honest about which of these applies. A unit whose actual problem is data preparation will not solve it by changing analysis tools.
The evaluation criteria
Analytical parity
Does the alternative support the traversals your analysts rely on — path finding, common neighbours, centrality, community detection, temporal filtering, geospatial overlay? Test on a real historical case, not a sample dataset.
Entity resolution quality
This is where products differ most and demonstrations reveal least. Load a dataset containing the same individuals under transliteration variants and inconsistent address formats and measure both false merges and missed merges. Check that every merge decision is visible, reversible and attributable.
Provenance and confidence
Can every entity and relationship carry a source, a date, an entering analyst and a confidence grade? Systems without this produce charts that cannot be defended in court.
Ingestion breadth
Test with your own files: multi-operator call records, several banks’ statement formats, a device extraction, a scanned document set, and a regional-language document. Measure analyst hours required per source type.
Deployment and sovereignty
Establish where processing happens, whether the system runs fully air-gapped, and what it transmits externally including licensing and telemetry. Verify with network capture during pilot.
Scale characteristics
Test at the largest dataset you realistically expect, not a representative sample. Chart rendering, query latency and entity resolution all degrade differently with volume.
Access control and audit
Role-based access aligned to case assignment, and query-level audit logging that cannot be altered by users.
Migration path
Can existing charts and entity data be imported? Can data be exported in an open, documented format if you later leave? Both directions matter, and the second is the one buyers forget.
Training and analyst transition
Analysts fluent in one idiom lose productivity when moved. Budget for it explicitly and evaluate the vendor’s training offering as a real cost line.
Support model and jurisdiction
Response times, language, on-site availability, and where support staff sit relative to your data.
Structuring the pilot
A meaningful evaluation runs a closed historical case end to end, with the analysts who will actually use the system, on your own data, on hardware you would actually procure. Score against the criteria above with the sovereignty and audit items treated as pass or fail.
Run the incumbent through the same pilot. Units frequently discover their dissatisfaction was with data preparation or training rather than with the analysis tool, which is a much cheaper problem to fix.
Where pi-labs fits
We build pi-scout, and we are not neutral about it. What we would say plainly is where it is a fit and where it is not.
It fits units whose binding constraints are deployment sovereignty, ingestion of Indian data formats, cross-case rather than within-case analysis, and multi-analyst collaboration on a shared dataset. It is built to run entirely inside agency infrastructure, including air-gapped, with provenance and query-level auditing as core rather than optional.
It is a less obvious choice for an organisation with a small number of analysts deeply invested in an existing desktop chart workflow, working case-sized datasets, with no sovereignty constraint. In that situation the transition cost may exceed the gain, and we would rather say so at the evaluation stage than after a procurement.
