Datadog BYOC Enhances Observability While Open AI Sharing Transforms SaaS Growth

Enterprise SaaS spent this week renegotiating its own boundaries. The most visible signal came from observability: Datadog, long associated with a classic multi-tenant SaaS delivery model, introduced a Bring Your Own Cloud (BYOC) option that lets customers run Datadog’s observability stack inside their own cloud environments [1]. In parallel, Databricks pushed on a different boundary—interoperability—by introducing OpenSharing, an open-source protocol aimed at sharing AI assets (models and agent skills) across domains and with external partners [2].

Meanwhile, the “AI inside the workflow” story continued to harden into product and program. Tempo expanded its AI roadmap with new Rovo Agents intended to increase automation in enterprise workflows, and paired that with an Enterprise Practitioner Program to help organizations implement these capabilities effectively [3]. And on the deployment side, Anthropic partnered with Tata Consultancy Services (TCS) to scale enterprise AI deployments, leaning on a services giant to accelerate integration of Anthropic’s models into enterprise environments [5].

Taken together, these moves point to a SaaS market that’s still growing—one forecast projects global SaaS reaching $1.52 trillion by 2035, driven by AI, cloud computing, and digital transformation [4]—but also maturing in how it delivers value. The week’s announcements weren’t just about new features; they were about where software runs, how AI artifacts move, and who does the heavy lifting to make “enterprise-ready” real.

Datadog BYOC: Observability steps outside the SaaS perimeter

Datadog’s BYOC model is a notable shift in SaaS delivery: enterprises can deploy Datadog’s observability tools within their own cloud environments rather than consuming them solely as a vendor-hosted service [1]. The immediate appeal is straightforward—greater control over data and a delivery model that can better align with internal policies around data handling and operational governance [1]. For organizations that have been cautious about sending telemetry, logs, and traces to a third-party SaaS endpoint, BYOC offers a different risk posture without necessarily abandoning the Datadog toolchain.

Why it matters: observability data is often among the most sensitive operational datasets an enterprise produces. It can reveal system topology, service dependencies, and incident patterns. BYOC is a direct response to customers that want the convenience and product velocity of a SaaS vendor, but with more control over where the data lives and how the platform is operated [1]. It also reflects a broader enterprise pattern: “cloud-first” doesn’t always mean “vendor-hosted,” especially when compliance, internal controls, or procurement constraints are involved.

The expert caution is equally important. Analysts noted that vendor lock-in can show up in multiple forms beyond data residency [1]. Even if telemetry stays in a customer’s cloud, lock-in can persist through proprietary query languages, dashboards, alerting semantics, integrations, and operational processes built around a specific platform. BYOC can reduce certain concerns, but it doesn’t automatically make observability portable.

Real-world impact: platform teams evaluating observability now have a more nuanced decision to make. The question is no longer only “which tool,” but “which operating model”—vendor-hosted SaaS, customer-hosted BYOC, or a mix. That choice will shape staffing, cost allocation, and incident response workflows as much as it shapes feature checklists [1].

Databricks OpenSharing: A protocol play for AI asset portability

Databricks introduced OpenSharing, described as an open-source protocol designed to enable sharing of AI assets—models and agent skills—across domains and with external partners [2]. The framing is explicitly about modernizing collaboration in AI development by promoting interoperability and reducing silos [2]. In a SaaS landscape where AI capabilities are increasingly embedded into platforms, the ability to move AI assets between teams, business units, and partner ecosystems becomes a strategic differentiator.

Why it matters: enterprises are increasingly building AI systems that span organizational boundaries—internal product teams, data teams, and external partners. When AI assets are trapped inside a single platform’s sharing mechanism, collaboration becomes brittle and slow. A protocol approach aims to make sharing more standardized, potentially lowering friction when organizations need to exchange models or agent skills across different environments [2].

The expert take here is less about a single feature and more about the direction: protocols tend to outlast product cycles. By positioning OpenSharing as open-source and protocol-based, Databricks is signaling that AI collaboration needs a layer that is not tightly coupled to one vendor’s UI or one vendor’s marketplace [2]. That’s a meaningful stance in a market where “AI platform” often implies a vertically integrated stack.

Real-world impact: if OpenSharing gains adoption, it could change how enterprises structure AI partnerships. Instead of negotiating bespoke integrations for each partner, teams could align on a common sharing mechanism for AI assets [2]. That could accelerate joint development, reduce duplication, and make it easier to operationalize AI across organizational boundaries—especially when agent skills and models need to be reused in multiple contexts.

Tempo’s Rovo Agents and practitioner program: AI features meet implementation reality

Tempo expanded its AI roadmap with new Rovo Agents aimed at enhancing automation within enterprise workflows [3]. Alongside the product direction, Tempo launched an Enterprise Practitioner Program intended to support organizations implementing these AI solutions effectively [3]. The pairing is telling: AI features alone don’t guarantee outcomes; enterprises often need enablement, patterns, and operational guidance to translate “agent” capabilities into reliable business processes.

Why it matters: workflow automation is where AI promises to become tangible—less about demos, more about repeatable execution. By focusing on agents designed for enterprise workflows, Tempo is targeting the operational layer where SaaS platforms can deliver measurable efficiency gains [3]. But the addition of a practitioner program suggests Tempo recognizes the adoption curve: governance, change management, and integration work can be the difference between a pilot and a scaled deployment.

Expert take: the most enterprise-ready AI roadmaps increasingly include not just product releases but also programs that help customers implement responsibly and consistently. Tempo’s Enterprise Practitioner Program is positioned as a support mechanism for implementation [3], which aligns with a broader trend of SaaS vendors building “customer operating systems” around AI—training, best practices, and structured adoption paths.

Real-world impact: for IT and business operations leaders, this is a reminder to evaluate AI announcements in two dimensions: capability and adoption support. Rovo Agents may expand what can be automated, while the practitioner program may reduce the time-to-value by helping teams deploy those capabilities in a controlled way [3].

Anthropic + TCS: Services-led scaling for enterprise AI deployments

Anthropic partnered with Tata Consultancy Services (TCS) to scale enterprise AI deployments, aiming to improve integration of Anthropic’s AI models into enterprise environments by leveraging TCS’s IT services and consulting experience [5]. This is a classic enterprise pattern: when a technology is powerful but complex to operationalize, services partners become the multiplier.

Why it matters: enterprise AI deployments often stall not because models are unavailable, but because integration is hard—connecting AI to systems of record, aligning with security and governance requirements, and embedding capabilities into workflows. A partnership with a major services firm is a direct route to scaling implementation capacity and standardizing delivery approaches across many customers [5].

Expert take: this move reinforces that “enterprise-ready AI” is as much about delivery as it is about model quality. By working with TCS, Anthropic is effectively acknowledging that large-scale adoption depends on the ecosystem that can implement, customize, and operate AI in real enterprise conditions [5].

Real-world impact: for buyers, the partnership can change procurement and rollout dynamics. Enterprises that already rely on large consultancies for transformation work may find it easier to adopt Anthropic’s models when integration and deployment are packaged into familiar services engagements [5]. For SaaS vendors more broadly, it’s a signal that channel and services strategy is becoming a core part of AI go-to-market.

Analysis & Implications: SaaS growth, but with new operating models and interoperability pressure

A forecast this week projected the global SaaS market will reach $1.52 trillion by 2035, growing at a 15.4% CAGR, driven by AI, cloud computing, and digital transformation [4]. The week’s announcements help explain what that growth could look like in practice: not simply “more SaaS,” but SaaS delivered through more varied architectures, with more emphasis on AI portability and implementation ecosystems.

First, Datadog’s BYOC underscores a shift from a single dominant delivery model to a spectrum. SaaS vendors are increasingly asked to meet enterprises where they are—sometimes that means vendor-hosted multi-tenant services, and sometimes it means running the vendor’s software inside the customer’s cloud environment [1]. This is not a retreat from SaaS value; it’s an attempt to preserve SaaS product velocity while accommodating enterprise control requirements. But the lock-in warning matters: BYOC can change data residency and operational control, yet still leave customers deeply dependent on a vendor’s semantics and tooling [1]. Enterprises should treat BYOC as an architectural option, not an automatic portability guarantee.

Second, Databricks’ OpenSharing points to a growing need for standardized ways to exchange AI assets across organizational boundaries [2]. As AI becomes embedded in SaaS platforms, the “unit of interoperability” is evolving—from data tables and APIs to models and agent skills. Protocols that reduce silos can become strategic infrastructure, especially for enterprises that collaborate with partners or operate across multiple internal domains [2].

Third, Tempo’s combination of Rovo Agents and an Enterprise Practitioner Program highlights the operational reality of AI adoption: enterprises need both capability and enablement [3]. This aligns with Anthropic’s partnership with TCS, which similarly emphasizes scaling deployment through services expertise [5]. In other words, the market is converging on a view that AI value is delivered through a blend of software, operating model, and implementation capacity.

The connective tissue across the week is maturity. SaaS is still expanding [4], but the competitive battleground is shifting toward control (BYOC), interoperability (OpenSharing), and scalable adoption (practitioner programs and services partnerships). Enterprises that align their platform strategy with these dimensions will be better positioned to capture AI-driven gains without accumulating unmanageable operational or vendor dependencies.

Conclusion

This week’s SaaS developments weren’t about incremental UI polish; they were about the shape of enterprise software delivery in an AI-heavy era. Datadog’s BYOC move suggests that “SaaS” is increasingly defined by product experience and update cadence, not strictly by where the software runs [1]. Databricks’ OpenSharing argues that AI collaboration needs open mechanisms for exchanging models and agent skills, especially as partnerships and cross-domain work become the norm [2]. Tempo and Anthropic, in different ways, reinforced that scaling AI is as much about programs and services as it is about models and features [3][5].

For enterprise leaders, the takeaway is to evaluate SaaS roadmaps through three lenses: deployment control, interoperability, and adoption capacity. The market may indeed be on a path toward trillion-dollar scale [4], but the winners—vendors and customers alike—will be those who treat architecture and implementation as first-class product concerns, not afterthoughts.

References

[1] Datadog observability breaks away from SaaS with BYOC — TechTarget, June 11, 2026, https://www.techtarget.com/news/it-management?utm_source=openai
[2] Databricks intros OpenSharing, a new standard for sharing AI — TechTarget, June 11, 2026, https://www.techtarget.com/news/it-management?utm_source=openai
[3] Tempo Expands AI Roadmap with New Rovo Agents and Enterprise Practitioner Program — ERP News, June 11, 2026, https://erpnews.com/?utm_source=openai
[4] SaaS Market Projected to Reach $1.5 Trillion by 2035, Fueled by AI and Cloud — SaaSRise, June 11, 2026, https://www.saasrise.com/news?utm_source=openai
[5] Anthropic taps TCS to scale its enterprise AI deployments — TechCrunch, June 11, 2026, https://techcrunch.com/category/enterprise/?utm_source=openai