Ashfaq Mohiuddin

Salesforce Acquires Informatica for $8B: A Strategic Move or a Step Backward? Salesforce’s recent $8 billion acquisition of Informatica has certainly made headlines. But is this move a visionary leap into the future or a nostalgic nod to the past? Imagine a horse trader acquiring a wagon maker to compete in the burgeoning automobile industry of the early 20th century. While both were titans in their time, the world was rapidly shifting beneath their feet. In today’s era of Generative and Agentic AI, agility and innovation are paramount. Nimble, AI-first startups are redefining the technological landscape, delivering unprecedented solutions at speed and scale. In contrast, legacy giants like Salesforce and Informatica, despite their storied histories, may find themselves encumbered by outdated infrastructures and slower innovation cycles. Generative AI is like the the Hulk: immensely powerful but requiring careful handling. To truly harness its potential, it must be paired with cutting-edge technologies that have evolved rapidly—technologies that differ vastly from those of the past three decades. Attempting to make one elephant dance is challenging enough; orchestrating a duet between two could be a comedic spectacle. Enterprises today seek the grace and agility of a trained ballerina, not the cumbersome movements of legacy behemoths. This acquisition prompts a critical question: Are we witnessing a strategic evolution or a nostalgic consolidation? The future will favor those who adapt swiftly, not those who cling to past glories. What do you think? https://lnkd.in/gPVXsnVY

17

2 Comments

No one has the right metadata for AI. Not yet. Everyone’s talking about Salesforce acquiring Informatica — and whether its metadata stack is what AI really needs. The hard truth: no one's cracked the playbook for enterprise AI apps that deliver fast, versatile value at scale — yet. We don’t know what the “right” metadata stack looks like. Instead of debating whose stack is better, we must define what good is — as we build. Here’s my practical test. Take this MCP record for a cart page view: "ai_context": { "actionable": true, "entity_types": ["user", "cart", "product"], "potential_operations": ["recommend_cross_sell", "detect_abandonment", "trigger_discount"], "anomaly_score": 0.02 } ...plus UI context, cart state, resource context, resource dependencies and action hints — all embedded in the same record. Can your metadata stack produce this — or is it built for lineage? Can it take an unprocessed log and enrich it for AI — 1️⃣ Programmatically? There won’t be time for humans in the loop. 2️⃣ Fast? Cart page views lose value within seconds. 3️⃣ Efficiently? AI needs millions of events to thrive 4️⃣ Adaptively? “Potential operations” must evolve as the system learns. 5️⃣ Nimbly? MCP is only six months old. Things will change. Most metadata stacks can't - yet. It’s an exciting problem to crack - together. Where are you investing today to make your metadata stack AI-ready?

77

11 Comments

The agent stack is taking shape—but there's a missing layer nobody's talking about yet. Arvind Jain at Glean published a thoughtful breakdown of the emerging agent architecture yesterday: Models, Orchestration, Context, Agents, Interfaces, Security. One line stood out: "In past data eras, we only focused on recording the decisions, not understanding how those decisions were made." Glean's answer? Context graphs—infrastructure that captures the topology of work, how decisions flow through an organization, what agents touched and when. That's a massive step forward. But it exposes a deeper question: When AI agents make consequential decisions at scale—pricing exceptions, legal risk assessments, hiring calls, capital allocation—what happens to the reasoning behind those decisions? Context graphs capture how work flows. Agent traces capture what was executed. But why alternatives were rejected, what criteria mattered, what assumptions shaped the judgment—that doesn't live in workflows or execution logs. Six months later, when: Boards audit AI decisions Regulators ask "why did the system decide this?" Successor teams inherit agent fleets Patterns of failure emerge The reasoning isn't retrievable. It evaporated the moment the decision was made. This is the gap between decision execution (what Glean's orchestration layer handles) and decision reasoning (what institutional memory requires). The agent stack needs a reasoning preservation layer. Not explainability in real-time (models are getting better at that). Not audit logs of what happened (SaaS already does that). Not workflow documentation (context graphs capture that). Infrastructure that captures judgment at the commit boundary—so reasoning survives retrieval when the consequences play out months or years later. That's what we're building at Quantum Mosaic. Not to replace any layer of the agent stack—but to add the reasoning preservation layer that makes institutional memory possible when agents make thousands of consequential decisions daily. Because as Arvind Jain wrote: understanding how decisions were made requires more than capturing what was done. It requires preserving why. Here is the original post from Arvind Jain: https://lnkd.in/ga2Cy76T

2

Commerce is being quietly reinvented by agents that actually know the customer. In a recent conversation with Gordon Evans, CMO of Commerce Cloud at Salesforce, we talked about what happens when AI agents are powered by real data—order history, inventory, location, preferences—not just prompts. His framing stuck with me: the future of shopping comes down to two things—conversation and context. And that applies inside the office just as much as it does in the cart. Full conversation on the Shift AI Podcast.

8

2 Comments

Enjoyed talking with Matthew Volm on RevOps AF about why most CRMs have an identity crisis. We dug into the “data behind the data” problem, the fact that 87% of CRM records have errors (based on data Kernel has fixed), and how that inaccuracy impacts territory design, account allocation, and rep productivity. I also went through the three-layer data stack: entity, enrichment and orchestration, and why getting the entity layer right is critical for your GTM strategy. If you want to avoid garbage-in, garbage-out and make your CRM ready for automation and AI, this episode is worth listening to. Link in comments.

29

2 Comments

Here is my take on Salesforce’s acquisition of Informatica: 1. 2. 3. That’s all the valuable things I have to say on this topic. #satire

41

7 Comments

I am pretty excited about what's going on at Snowflake particularly around automating the front end of data science and analytics. Check out my latest short on the topic below.

4