Resources
Case studies without invented numbers.
Why this page is not a wall of fake logos
Enterprise buyers are used to case-study pages stuffed with anonymous “40% efficiency gains” and logos nobody authorised. That pattern trains distrust. We refuse it.
Named narratives require client permission, careful scoping of what can be disclosed, and enough substance to be useful — problem, approach, constraints, and qualitative outcomes. Until we have that package, we will not invent one to look thick for search engines.
What you get instead: a clear publishing standard, a reference process you can use today, and the organisation names we already use publicly elsewhere on this site.
What a published Arcloops case study will include
- Context — sector, organisation shape, and the constraint that triggered the work (without leaking confidential detail).
- Starting point — readiness gaps, shadow tools, or stalled pilots — stated honestly.
- Approach — which loops we used (audit, consult, train, build) and why that sequence fit.
- What shipped — systems, programmes, or policy artefacts the client owns after handover.
- Outcomes we can stand behind — qualitative results and only metrics the client has approved in writing. Never invented percentages.
- Limits — what we would do differently, and what the engagement did not solve.
Engagement patterns we discuss on reference calls
Without publishing a full narrative, we can still describe patterns from work we have done — useful for buyers who need to know we have sat in similar rooms:
Readiness before spend
Organisations under board pressure to “do AI” that needed an evidenced baseline before buying tools — data, process owners, literacy, and shadow footprint mapped in days, not quarters.
Department workflows that hand over
HR, finance, procurement, and service teams that needed systems shaped to Bangla/English reality and existing stacks — not another ChatGPT licence pretending to be a programme.
Regulated controls
Banks, NBFIs, and adjacent firms that needed policy, governance, and vendor discipline so production AI could survive scrutiny — without freezing every experiment.
Industrial and RMG constraints
Operations where floor supervisors, buyer audits, and messy master data decide whether a pilot dies — change and training treated as part of delivery.
How to request a relevant reference
Tell us your industry, organisation size band, and the constraint you are trying to solve. We will say whether we have a conversation we can stand behind — and whether the peer is willing to talk under NDA.
We will not invent a reference. If we cannot match your sector cleanly, we will say so and point you to insights or an audit instead of forcing a bad fit.
Organisations we’ve worked with
Names we already use publicly. No fabricated logos, quotes, or ROI attached to these names on this page.
- Grameenphone
- SMC
- Sajida Foundation
- LGED
- Krayons
Case studies FAQ
When clients grant named permission and we can write a narrative that meets our standard. We will not ship thin stubs with placeholder metrics to fill the sitemap.
Often yes — subject to the peer organisation’s willingness. Contact us with your sector and constraint; we will confirm what is possible.
Because invented or unapproved numbers destroy trust. When a client approves a metric in writing, it can appear in a narrative. Until then, we describe patterns and process.
Yes. /work permanently redirects here. The honesty standard is intentional — not a missing page.
Need a relevant reference?
Tell us your industry and constraint. We’ll connect you to a conversation we can stand behind — not a deck with invented numbers.